The coverage of residm?

chenmy · Post by **chenmy** » Tue Sep 09, 2014 8:22 pm

Hello,
There are three question regarding the Ba(Mg1/3Ta2/3)O3 gamma phonon calculation, which were applied by LDA-HGH pseudopotential. The material behave as wide gap insulator(diemac=4).
Electron numbers of Ba, Mg, Ta, O are 10,2,5, and 6. The total electrons of unit cell are 96 electrons. Default nband number was 50,but we need to calculate phonon dispersion curve, nband was chosen as 64((nkpt* nsppol)).
Q1: Before the gamma point phonon calcaultion, fully relaxed structure was needed. The relative parameters were tolmxf=1*10^-6, toldff=1*10^-8.
The output file showed the small stress tensor (10^-9-10^-8)and retained same symmetry.
Then gamma point calculation had two dataset: GS and phonon calculation. we applied the crystal parameters of relaxed structure and wfk file into GS calculation(toldff=1*10^-8).
The parameters of output in dataset1(GS) reached residm~10^-6, vres2~10^-13,diffor~10^-9, maxfor~10^-5.
However, the output stress tensor in GS calculation(10^-4) were much higher than that in relaxation calculation(10^-9-10^-8).
Is it reasonable? or anything wrong in my calculation?

Q2: continued Q1, after obtaining wfk from dataset1(GS), phonon of gamma point were calcaulted(tolvrs=1*10^-10, nkpt=32).
However, when the coverage was reached(tolvrs2<10^-10), residm values show 10^-1.
This situation in phonon calculation of non-gamma point become worse(10^-1-10^+5).
Is it acceptable or anything wrong in my calculation?

Q3: The 3 acoustic phonon in this calculation were not zero, and around 20 cm-1.
Does it mean the error bar are 20cm-1 and I have to change my pseudopotential?

all relative parameter were showed in attached output file. please help me find the answers of these problems,
Thanks a lot,

MeiYu